Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 DOI 10.1186/s12974-017-0855-0

RESEARCH Open Access activation is essential for BMP7- mediated retinal reactive gliosis Subramanian Dharmarajan1,2, Debra L. Fisk3, Christine M. Sorenson4, Nader Sheibani3 and Teri L. Belecky-Adams1,2*

Abstract Background: Our previous studies have shown that BMP7 is able to trigger activation of retinal macroglia. However, these studies showed the responsiveness of Müller glial cells and retinal in vitro was attenuated in comparison to those in vivo, indicating other retinal cell types may be mediating the response of the macroglial cells to BMP7. In this study, we test the hypothesis that BMP7-mediated gliosis is the result of inflammatory signaling from retinal microglia. Methods: Adult mice were injected intravitreally with BMP7 and eyes harvested 1, 3, or 7 days postinjection. Some mice were treated with PLX5622 (PLX) to ablate microglia and were subsequently injected with control or BMP7. Processed tissue was analyzed via immunofluorescence, RT-qPCR, or ELISA. In addition, cultures of retinal microglia were treated with vehicle, lipopolysaccharide, or BMP7 to determine the effects of BMP7-isolated cells. Results: Mice injected with BMP7 showed regulation of various inflammatory markers at the RNA level, as well as changes in microglial morphology. Isolated retinal microglia also showed an upregulation of BMP-signaling components following treatment. In vitro treatment of retinal astrocytes with conditioned media from activated microglia upregulated RNA levels of gliosis markers. In the absence of microglia, the mouse retina showed a subdued gliosis and inflammatory response when exposed to BMP7. Conclusions: Gliosis resulting from BMP7 is mediated through an inflammatory response from retinal microglia. Keywords: Microglia, Reactive gliosis, BMP7, Retina, Müller , Retinal astrocytes

Background contain and another glial-like cell type, The mammalian retina consists of at least two known as the non-astrocytic inner retinal glia-like (NIRG) distinct glial populations: the macroglia, which in- cells, that reside in the INL of the chick retina [4, 5]. cludes Müller glia and retinal astrocytes, and the Müller glial cells and retinal astrocytes are essential for microglia. The Müller glia are the primary glial cells maintaining retinal . Any injury or disease found in the retina, having their nucleus in the inner leading to retinal damage or disruption of the homeosta- nuclear layer (INL) with processes extending from the sis triggers the glial cells to become active, a response inner limiting membrane at the vitreal border to the termed reactive gliosis. Reactive gliosis has been outer limiting membrane at the base of the photo- observed in all retinal disease and injury models includ- receptor inner segments [1]. Retinal astrocytes ing glaucoma, age-related macular degeneration, and migrate into the retina from the optic nerve and diabetic retinopathy [6–9]. Reactive gliosis is character- reside in the nerve fiber layer [2]. The microglia are ized by hypertrophy, altered function brought about by the resident found scattered through all changes in expression of proteins such as glutamine the retinal layers [3]. The retina of some species also synthetase (GS), S100-β, proteins, chondroitin sulfate (CSPG), matrix metal- loproteinases (MMP), and an increase in growth factors * Correspondence: [email protected] 1Department of Biology, Indiana University-Purdue University Indianapolis, such as ciliary neurotrophic factor (CNTF), leukemia 723 W Michigan St, SL306, Indianapolis, IN 46202, USA inhibitory factor (LIF), and vascular endothelial growth 2 Center for Developmental and Regenerative Biology, Indiana factor (VEGF) [10, 11]. Multiple factors can trigger University-Purdue University Indianapolis, 723 W Michigan St, Indianapolis, IN 46202, USA Full list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 2 of 18

gliosis, including the bone morphogenetic proteins stimulating factor (CSF) and VEGF, and various scavenger (BMPs) [12–14]. Recent evidence from the Belecky- receptors and -presenting molecules such as the Adams laboratory showed that BMP7 triggered gliosis in scavenger receptor A (SR-A) and major histocompatibility both the Müller glia and astrocytes of the mouse retina; complex (MHC) [3, 27]. Furthermore, research has however, the mechanism by which BMP7 triggers gliosis revealed that activated microglia can be further classified is unknown [11]. into the following phenotypes: the M1 or proinflammatory The BMPs are growth factors that belong to the trans- phenotype and the M2 or the anti-inflammatory pheno- forming growth factor beta (TGF-β) superfamily. BMP type [28, 29]. Polarization to the M1 phenotype, following signaling is initiated following the binding of the ligand exposure to factors such as lipopolysaccharide (LPS) and to serine threonine kinase receptors. This leads to the IFN-γ, the microglia upregulate proinflammatory factors activation of the receptors and the subsequent phos- such as IL-1β, tumor factor alpha (TNF-α), phorylation and activation of downstream signaling inducible synthase (iNOS), SRs, and MHC-II components. In the canonical pathway, the BMP signals [30, 31]. The M2 phenotype plays a role in the resolution by phosphorylation and activation of downstream recep- of the inflammation and tissue remodeling. This pheno- tor SMADs (RSMADs). The RSMADs form a dimer type is induced by factors such as IL-4 and IL-10 or with the co-SMAD (SMAD4) and are shuttled to the through the maturation of the M1 cells. This phenotype nucleus to regulate transcription. BMP can also mediate was characterized by an upregulation of markers such as the activation of a non-canonical pathway referred to as arginase-1 (Arg-1) and mannose receptor (Mr), the BMP mitogen-activated protein kinase pathway such as IL-10 and IL-13, and growth factors such as (BMP-MAPK). In the BMP-MAPK pathway, the receptors TGF-β and VEGF [30, 32]. recruit the X-linked inhibitor of (XIAP) to a Signals from neurons and macroglia, such as fractalk- complex containing TAB1 and TAK1, thereby activating ine, , and neurotrophins help keep the TAK1. TAK1 then activates downstream kinases, eventu- glial population in the quiescent state [6, 33]. Activation ally activating NF-κB, p38, and JNK MAPKs [15, 16]. In of the glial cells has been found to be mediated by the CNS, BMP regulation has been observed in various similar stimuli in vitro and in retinal disease models in diseases and injury models, such as injuries, vivo [6, 25, 34–36]. Cytokines and other inflammatory axonal damage, and [14, 17, 18]. In the retina, markers such as TNF-α, iNOS, CNTF, and LIF are not upregulation of BMPs and their signaling components are only regulated during gliosis but are also factors known observed in the photo-damaged retina injury model and to act on the glial cells and regulate gliosis [20, 37, 38]. in diabetic retinopathy [19–21]. Activated microglia are known to regulate Müller cell Microglia are the innate immune cells of the retina. In activity directly, regulating cell morphology, prolifera- their resting state, the microglia act as sentinels, extend- tion, and gene expression [26, 39]. Activated microglia ing their processes throughout the retina. In the mouse can also regulate the generation of Müller glia-derived retina, the microglia are initially found in the ganglion progenitors [40]. Here, we provide evidence that cell layer, entering the retina from the ciliary marginal supports the hypothesis that BMP7 indirectly triggers zone and vitreous. By postnatal day 7, the microglia gliosis by activating the proinflammatory state of retinal spread to the rest of the retinal layers, finally resting in microglia. the plexiform layers [22]. Upon receiving signals from injured or dying cells, the microglial cells become acti- Methods vated: they retract their processes, undergo an increase Cell culture in cellular area, become amoeboid in shape, and migrate Mouse retinal astrocytes were isolated in the Sheibani to the area of injury or disease to phagocytize cellular lab and maintained as previously described in [11, 41]. debris and metabolic products [23, 24]. Stimuli such as Microglial cells were isolated from retinas of newborn neuronal loss or damage, inflammation, and nerve (P0-P4) immortomouse back crossed into C57BL/6J as degeneration activate the microglia into a motile effector described in [42] with some modifications. Briefly, the cell with altered morphological characteristics [25, 26]. retinas were placed in a solution of Trypsin/EDTA Microglial activation has been observed in all retinal (5 ml; 0.25% trypsin and 1 mM EDTA; Thermo diseases, including diabetic retinopathy, age-related Scientific) and incubated at 37 °C for 5 min. Following macular degeneration, glaucoma, and models of retinal incubation, the samples were triturated by pipette, and pathologies. In addition to the morphological changes 5 ml of DMEM with 10% FBS was added to stop trypsin following activation, microglia also induce a change in activity. The digested tissue was centrifuged for 5 min production of various cytokines such as interleukin 1 beta at 400 × g at room temperature, the supernatant was (IL-1β), IL-6, and (IFN-γ), chemokines carefully aspirated, and the pellet was re-suspended in such as RANTES, MCP1, growth factors such as colony the microglia medium [a 1:1 mixture of DMEM: F12 Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 3 of 18

(Thermo Scientific) containing 10% FBS and 44 U/ml with the vehicle control in the left eye and BMP7 in of interferon-γ (R&D Systems, Minneapolis, MN)], the right eye. P30 VE-YFP mice (n =3),whichexpress plated on a single well of a 6-well plate, and incubated YFP in endothelial cells, generated by crossing a line in a tissue culture incubator at 33 °C and 5% CO2.The of mice containing an enhanced yellow fluorescent cells were allowed to grow for 1-2 weeks and fed every protein(YFP)withafloxedstopsequenceupstream 3-4 days until nearly confluent. The medium was then of the YFP (B6.129X1-Gt(ROSA)26Sortm1(EYFP)Cos/J; strain removed from the plate and rinsed with PBS containing number 006148 Jackson laboratory) [44] with the VE-cad- 0.04% EDTA. The plate was then incubated with 2 ml herin-cre line (B6.FVB-Tg(Cdh5-cre)7Mlia/J; stock num- of PBS containing 0.04% EDTA and placed on a multi- ber 006137 Jackson laboratory) [45], were used for purpose rotator at 100 rpm at room temperature for immunofluorescence experiments determining PU.1 20–30 min. The supernatant was collected in a 15-ml co-localization in the retina. tube containing 3 ml of DMEM with 10% FBS and cen- trifuged at 400 × g for 5 min. The detached cells were then re-plated in the microglia medium, allowed to Intraocular injections reach confluence, and expanded into 60-mm dishes. Postnatal day 30 (P30), C57BL/6J mice were anesthetized The purity of the microglial cultures was inspected by with ketamine and xylazine cocktail. Mice were injected immunocytochemical staining and flowcytometric ana- intravitreally with 1 μl of vehicle (4 mM HCL with lysis for F4/80 (eBiosciences; San Diego, CA) and kera- 0.1%BSA) or 1 μl BMP7 (20 ng/μl) as previously stated tin sulfate (Seikagaku Corporation; Jersey City, NJ). The in [11]. Intraocular injections were performed using a purity of culture was nearly 95% using FACS and im- manual microsyringe (World Precision Instruments) and munostaining analysis. Astrocytes and microglia were pulled glass micropipettes. grown in tissue culture dishes (BD Falcon) in an incu- bator with 5% CO2 at 33 °C and passaged every 5– 7 days using trypsin-EDTA, and the medium changed Microglia ablation every 3–4 days. Cells were treated with 1 μl/ml vehicle C57BL/6J mice were kept on chow feed containing (4 mM HCL with 0.1% BSA), 100 ng/ml of mouse bone 1200 ppm PLX5622 (PLX; Plexxikon Inc.) for up to morphogenetic protein 7 (BMP7; R&D systems), 21 days, starting at P30. Eyes were harvested at 7, 14, 300 ng/ml mouse interferon-gamma (IFN-γ;R&Dsys- and 21 days following start of the PLX diet for assess- tems), or 100 ng/ml LPS (Sigma). Medium from micro- ment of loss of microglia. The control mice were kept glial cells incubated with BMP7 or vehicle (conditioned on the control chow supplied by Plexxikon Inc. To medium) for 24 h was used to treat retinal astrocytes. determine if loss of microglial cells affected BMP7- The conditioned medium was added to the retinal as- mediated gliosis, mice were maintained on PLX chow trocytes medium at 25% concentration in the presence for the entirety of the experiment. In some animals, eyes of DMSO or 2.5 μM ALK2/ALK3/ALK6 inhibitor were injected with 1 μl vehicle (4 mM HCL with LDN193189 [43]. The retinal astrocytes were allowed 0.1%BSA) or 1 μl (20 ng/μl) BMP7 14 days following to grow for 24 h; after which, cells were harvested for treatment with PLX, and eyes were harvested and RNA isolation and RT-qPCR analysis. processed 3 and 7 days postinjection.

Experimental groups Tissue processing Experiments were carried out in 4–8 weeks old male Eyes from euthanized C57BL/6J mice were enucleated, C57BL/6J. All procedures were in accordance with washed in PBS, and either fixed in 4% paraformalde- the guidelines set by the Institutional Animal Care hyde (PFA) for immunofluorescence (IF) or dissected and Use Committee (IACUC) at the school of science to isolate the retina for preparation of RNA and/or IUPUI (protocol number SC230R). For BMP7 injec- protein. For IF analysis, enucleated eyes were washed tion studies, n = 8 mice were used with the left eye and fixed in 4% PFA, incubated in ascending series of injected with the vehicle and the right eye injected sucrose, and frozen in a sucrose OCT solution as with BMP7. For the PLX studies, two groups of mice previously described [11]. Thick sections (12 μm) were considered, the age-matched control chow group were cut using Leica CM3050S cryostat onto Super- (n =12)andthePLXgroup(n = 12), kept on PLX frost Plus slides (ThermoScientific) and stored at −80 °C chow diet. For the PLX BMP7 injection studies, two until use. Retinas from enucleated eyes were isolated groups of mice were considered, the age-matched as previously described [11]. Isolated retinas were control group (n = 12) and the PLX group (n =12), immediately processed for RNA isolation using kept on the PLX diet. Both the groups were injected RNeasy kit (Qiagen). Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 4 of 18

RT-qPCR assessing changes in inflammation are listed in Reverse transcriptase-quantitative polymerase chain Table 1. RT-qPCR was performed using SYBR green reaction (RT-qPCR) was performed to detect changes in master mix (Roche) with the reactions carried out in markers associated with gliosis and inflammation as the LighCycler480 system (Roche). The change in −ΔΔ previously described [11]. The primers for RT-qPCR RNA levels was measured using the 2 Ct method, analysis are listed in Table 1 of [11]. Included in this where Ct is the crossing threshold/crossing point table is the accession number of each gene, the (Cp) value. Relative RNA levels were calculated using sequence of each primer, product length, and calcu- the geometric means from the Ct value derived from lated efficiency of each primer. Primers used for three housekeeping genes: β- 2 Microglobulin (B2m), succinate dehydrogenase complex subunit A (Sdha), and signal recognition particle 14 kDa (Srp14).Ano template control was also tested for each marker. Table 1 The primers used for qPCR analysis Gene Primer Sequence Product length Gm-Csf Forward AGTCGTCTCTAACGAGTTCTCC 178 Immunofluorescence Reverse AACTTGTGTTTCACAGTCCGTT Frozen tissue sections were labeled as previously described [11]. Antigen retrieval was performed by using Csf1 Forward ACCAAGAACTGCAACAACAGC 91 1% sodium dodecyl sulfate (SDS) in 0.01 M PBS (5 min at Reverse GGGTGGCTTTAGGGTACAGG room temperature) or by heat antigen retrieval method. Inf-α Forward CAAGCCATCCCTGTCCTGAG 131 Briefly, sections were washed with 1× PBS, postfixed with Reverse TCATTGAGCTGCTGGTGGAG 4% PFA, and permeabilized with methanol. Sections were Inf-γ Forward CAACAGCAAGGCGAAAAAGGA 90 then incubated in 10 mM sodium citrate buffer at 65 °C Reverse AGCTCATTGAATGCTTGGCG for 45 min, allowed to cool at room temperature (RT) for 20 min, rinsed in deionized (DI) water 3×, and washed in Il-1β Forward TGTCTGAAGCAGCTATGGCAA 141 PBS once. To reduce autofluorescence, slides were then Reverse GACAGCCCAGGTCAAAGGTT incubated with 1% sodium borohydride in PBS for 2 mins Il6 Forward ACTTCACAAGTCGGAGGCTT 111 at RT. Slides were then blocked with 10% serum (goat or Reverse TGCAAGTGCATCATCGTTGT donkey) in 1× PBS with 0.25% TritonX-100 for 1 h VEGF Forward ACTGGACCCTGGCTTTACTG 74 followed by primary antibody diluted in blocking buffer Reverse CTCTCCTTCTGTCGTGGGTG overnight at 4 °C. Slides were then incubated with Dylight conjugated secondary antibodies (1:800; Jackson Immu- Tnf-α Forward TAGCCCACGTCGTAGCAAAC 136 noresearch) or Alexa flour (1:500; Invitrogen) conjugated Reverse ACAAGGTACAACCCATCGGC secondary antibodies for 1 h at RT in the dark, washed Ccl5 Forward TGCCCACGTCAAGGAGTATTT 111 with 1× PBS, incubated with Hoechst staining solution Reverse ACCCACTTCTTCTCTGGGTTG (2 μg/ml in PBS), and then mounted with Aqua Thbs1 Forward GCCACAGTTCCTGATGGTGA 149 Polymount (Polysciences). Reverse TTGAGGCTGTCACAGGAACG Biotin-streptavidin amplification was done by incubating slides with biotinylated antibody (1:500; Thbs2 Forward GGGAGGACTCAGACCTGGAT 105 Vector Labs) for 1 h at RT followed by Dylight conju- Reverse CGGAATTTGGCAGTTTGGGG gated to streptavidin (1:100; Vector Labs) for 1 h at Cd45 Forward TGACCATGGGTTTGTGGCTC 134 RT, in lieu of Dylight or Alexa flour conjugated Reverse TTGAGGCAGAAGAAGGGCAT secondary antibodies. For co-labeling involving pri- Cd68 Forward AAGGGGGCTCTTGGGAACTA 139 mary antibodies made in the same host, tyramide Reverse AAGCCCTCTTTAAGCCCCAC signal amplification was performed as per manufac- turer’s protocol (Perkin Elmer). For primary anti- Iba1 Forward ACGAACCCTCTGATGTGGTC 118 bodies made in mouse, reagents from the mouse on Reverse TGAGGAGGACTGGCTGACTT mouse kit (Vector Labs) were used for blocking and Irf8 Forward CGGATATGCCGCCTATGACA 73 primary antibody dilution. Labeled slides were imaged Reverse CTTGCCCCCGTAGTAGAAGC using Olympus Fluoview FV 1000. Antibodies used Sdha Forward GGACAGGCCACTCACTCTTAC 130 for immunofluorescence are listed in Table 2. Cell Reverse CACAGTGCAATGACACCACG counts were performed using the cell counter plugin of ImageJ. 40× images (n = 9) of retinal sections la- Srp14 Forward CCTCGAGCCCGCAGAAAA 134 beled with SOX9, CALBINDIN, CHX10, and BRN3A Reverse CGTCCATGTTGGCTCTCAGT and 60× image of retinal flatmounts (n = 8) labeled Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 5 of 18

Table 2 List of antibodies Marker Company If concentration Flatmount concentration Western blot concentration IBA1 WAKO 1:500 1:250 PU.1 CELL SIGNALING 1:100 SOX9 MILLIPORE 1:500 GαT1 SANTA CRUZ 1:100 CALBINDIN SIGMA 1:250 CHX10 EXALPHA 1:500 BRN3A CHEMICON 1:250 GFAP DAKO (polyclonal) 1:250 1:100 GFAP DAKO (monoclonal) 1:1000 S100β ABCAM 1:300 1:1000 NCAN R&D SYSTEMS 1:100 TXNIP SANTA CRUZ 1:250 phospho-SMAD1/5/9 CELL SIGNALING 1:100 phsopho-TAK ABCAM 1:500 β - TUBULIN SIGMA 1:1000 GFP THERMO SCIENTIFIC 1:150 for GαTRANSDUCIN were used for cell count ELISA analysis. Retinal thickness was measured on cross sec- Enzyme linked immunosorbent assay (ELISA) for IFN-γ tion of retina 200 μmawayfromtheopticnerve. was performed on media from treated cells in vitro or from whole mouse retina protein lysates using the mouse IFN-γ ELISA kit (Cat # ENEM1001, ThermoScientific) as Western Blotting per manufacturer’sprotocol. Extraction of proteins from retinal tissue was performed using lysis buffer as previously described in [11]. Briefly, Retinal flatmounts retinal tissue was homogenized in PBS and centrifuged at Preparation of retinal flatmounts and immunolabeling 13,000 rpm, 4 °C for 10 min. The supernatant was was done as described in [46]. Briefly, enucleated eyes discarded, and the pellet incubated with lysis buffer were washed in 1× PBS, fixed in 4% PFA for 15 min, (150 mM NaCl, 50mMTris pH 8.0, 2 mM EDTA, 5% transferred to 2× PBS on ice for 10 min, and followed by TritonX-100; 100 mM PMSF and protease inhibitor retina isolation. Four to five radial incisions were made cocktail, RPI corp.) for 20 min at 4 °C. The samples were in the retina to create a petal shape. Excess PBS was centrifuged at 13,000 rpm, 4 °C for 10 min, and total absorbed, and retinas were transferred to cold methanol protein was estimated using BCA protein assay kit (−20 °C) for 20 min. The tissue was washed with 1× PBS (ThermoScientific). and blocked in Perm/Block solution (1× PBS, 0.3% Forty micrograms of protein was loaded onto 4–20% TritonX-100, 0.2% bovine serum albumin, and 5% SDS precast gels (Expedeon), placed in a Biorad gel run donkey or goat serum). The tissue was then washed in apparatus, and run at 150 V for 1 h. Proteins were trans- PBSTX (1× PBS + 0.3% TritonX-100) and incubated with ferred onto a PVDF membrane, which was blocked with a primary antibody (Table 2) overnight at 4 °C. On the 5% milk in tris buffered saline tween 20 (TBST) at RT for following day, the tissue was washed in PBSTX, incu- 1 h on a shaker. The blots were incubated with primary bated with secondary antibody, washed, incubated with antibody-diluted TBST at 4 °C overnight on a shaker. The Hoechst solution, and mounted onto a slide with Aqua following day, the blots were washed in TBST and Polymount (Polysciences, Inc). Labeled slides were incubated with HRP conjugated secondary diluted 1:5000 imaged using Olympus Fluoview FV 1000. Morpho- in TBST for 1 h at RT. Blots were washed in TBST, logical analysis of labeled microglia (n = 4 per time incubated with super signal west femto chemilumines- point) for changes in area and number of branches was cent substrate (ThermoScientific), and visualized on performed using the Scholl analysis plugin in Fiji image x-ray films. β-TUBULIN was used as a loading analysis software [47]. Briefly, the flatmount image was control, and the concentrations of antibodies used are loaded on to the Fiji software and converted to binary. listed in Table 2. To calculate the area of the cell, the “Measure” plugin Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 6 of 18

was selected from the “Analyze” options. To determine Results the number of branches, a center of analysis was BMP signaling in retinal microglia defined via the straight line method. This line was Previous studies have shown that BMP7 triggers reactive drawn from the center of the cell to the end of the lon- gliosis of the retinal macroglia. Both the canonical as well gest branch to define a valid “Startup ROI.” The pro- as the non-canonical BMP-MAPK pathways were active gram was run on the default parameters with the in the retinal Müller cells and astrocytes following BMP7 starting radius set at 10 pixels. treatment [11]. However, the mechanism by which BMP7 triggered gliosis remains unclear. To determine if any of these pathways were activated in the microglia of control- Statistical analysis or BMP7-treated retina, double-label immunohistochem- Statistical analysis was performed via unpaired Students’s t istry was performed using antibodies to phospho SMAD test using SPSS software (IBM) between control/vehicle 1/5/9 (pSMAD), phospho TAK1 (pTAK1), and PU.1 (a and treated groups for RT-qPCR, cell counts, and microglia nuclear marker of microglia; Additional file 1: Figure S1) morphology. RT-qPCR and densitometries from PLX and on adult retinas following intravitreal injection of vehicle control mice injected with vehicle or BMP7 were analyzed or BMP7. In both vehicle- (Fig. 1a–d) and BMP7-treated viaonewayANOVAwithTukey’s test for post hoc analysis. retinas (Fig. 1e–h), sections showed nuclear co-labeling p ≤ 0.05 were considered to be statistically significant. with PU.1 and pSMAD. In contrast, pTAK1 was localized

Fig. 1 pSMAD and pTAK1 are localized to retinal microglia. Retinal sections from P30 mouse injected with vehicle or BMP7 24 h postinjection were double-labeled with antibodies that label microglial nuclei (PU.1) and phospho SMAD 1/5/9 (pSMAD; a–h) or phospho TAK1 (pTAK1; i–p). Thin plane confocal microscopy images with y,z (strips to right of the panel) and x,z planes (strips at the bottom of the panels) shown in (d), (h), (l), and (p). pSMAD-labeled cells were primarily found in the GCL in the vehicle-treated retina, with some colocalization with the nuclear microglial marker PU.1 (a–d). The BMP7-injected retina had an increase in pSMAD expression in the INL as well as substantial colocalization with PU.1 (e–h). In contrast, vehicle-injected retina showed pTAK1 expression in the GCL with little to no PU.1 colocalization (i–l), while the BMP7-injected retinas showed increased levels of pTAK1 levels in the INL, as well as significant colocalization with PU.1 (m–p). Magnification bar in a =50μm, for images (a–p) Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 7 of 18

primarily to the nuclei of GCL of vehicle-injected retinas an ELISA using medium from microglial cells incu- with no apparent co-localization with PU.1 (Fig. 1i–l), but bated with BMP7 for 24 h and whole retinal lysates co-labeled PU.1+ cells in the BMP7-treated retinas, in from mice treated with vehicle or BMP7 (Fig. 2d). addition to other cells in the INL and GCL (Fig. 1m–p). Values plotted in graph are relative to the respective There was also a striking increase in the localization vehicle controls; hence, increases in mRNA levels in com- of pTAK1 in both the inner and outer plexiform parison to controls are bars above a level of 1.0, while bars layers of BMP7-treated retinas that was not apparent below the level of 1.0 represent a decrease. We observed a in vehicle-treated retinas (Fig. 1m–p). Retinal sections 2-fold increase in the IFN-γ protein levels in the astrocytes were also co-labeled with IBA1 and pTAK1 or and microglial cell medium, and a 5-fold increase in IFN-γ pSMAD to show localization in microglia (Additional protein level was detected in retinal lysates 7 days file 2: Figure S2). Negative controls showed no label posttreatment with BMP7 compared with vehicle. (Additional file 3: Figure S3). Changes in morphological characteristics of microglia following control and BMP7 treatments were subse- BMP7 induces inflammatory changes in vivo quently investigated. It has been reported by other To determine whether BMP7 regulated inflammatory investigators that activated microglia increase in area signals that could then either trigger or enhance the with an increase in branch points [48]. Retinal flat- gliosis response, BMP7-treated retinas were analyzed for mounts of 1 day BMP7- and vehicle-treated retinas were messenger RNA (mRNA) levels of proinflammatory labeled with IBA1 and analyzed for average cell area and markers (Fig. 2a). For the analyses of mRNA levels, number of branch points in cellular processes (Fig. 3a, values plotted in graphs were all relative to control levels b). Graphs show relative changes in the area and number which were set to a value of 1.0; hence, increases in of branches (“Median intersections” output from the mRNA levels in comparison to controls are bars above a Sholl analysis). Morphological analysis revealed that the level of 1.0, while a decrease is represented by bars BMP7-treated retinas contained microglia with a larger below the level of 1.0. Three days postinjection, area in comparison to vehicle-treated retinas, and a increases of 1.5-fold or more in mRNA levels of Tnf-α, decrease in the number of branches (Fig. 3c). Il-1β, and Ifn-γ were present. However, larger increases were evident in multiple factors 7 days postinjection, Activated microglia secrete factors that induce gliosis including granulocyte colony stimulating We have observed that BMP7 is able to activate retinal factor (Gm-Csf), colony stimulating factor (Csf), Ifn-α, microglia in vitro and in vivo (Figs. 2 and 3, respect- Ifn-γ, Il-6, Vegf, thrombospondins-1 and-2 (Thbs1 & ively). To determine if microglia secrete factors that Thbs2), and Cd68. We also observed more than a 2-fold trigger reactive gliosis in vitro, we used conditioned increase in microglial marker Iba1 and Irf8, markers for medium obtained from mouse microglia cultures activated microglia. treated with vehicle (vehicle conditioned media) or To determine if the increases in proinflammatory BMP7 (BMP7-conditioned media) for 24 h, and used markers present in BMP7-treated retinas were medi- for treatment of mouse retinal astrocytes (Fig. 4b–d). ated by retinal microglial cells, the effect of BMP7 Graphs represent mRNA levels in cultures treatment on isolated mouse retinal microglial cells in treated with BMP7-conditioned media relative to vitro was observed using RT-qPCR. mRNA levels cultures treated with vehicle-conditioned media; were investigated in microglial cells incubated with pretreated with DMSO or LDN193189. Retinal astro- vehicle or BMP7 for 3, 6, 12, or 24 h (Fig. 2b). Again, cyte cells were incubated for 24 h with microglial cell-- changes in mRNA levels relative to controls were conditioned medium and were assessed for changes in plotted, where a value of 1.0 indicates levels of con- markers associated with gliosis. To reduce the possibil- trol mRNA. Following 3 h of incubation with BMP7, ity that the BMP7 added to the microglial medium only levels of Ifn-γ were 1.5-fold greater, whereas at might directly affect the astrocytes, an inhibitor of BMP 6 h the average mRNA levels of Gm-csf, Ifn-γ, Csf1, Tnf-α receptors, LDN193189, was added to the conditioned and Il-6,andCd68 were increased to 1.5-fold above con- medium (Fig. 4d). RT-qPCR analysis showed a statisti- trol or greater (Fig. 2b). By 24 h of incubation, many cally significant increase in expression of gliosis ofthemoleculelevelsweredecreasedincomparison markers glial fibrillary acidic protein (Gfap), S100-β, to the 6-h time point; however, Ifn-γ and Thbs2 were Gs, epidermal growth factor receptor (Egfr), and phos- increased in comparison to control and 6-h mRNA phacan (Pcan) 1.5-fold above that of astrocyte cells levels. As a positive control for inflammation, micro- treated with DMSO and vehicle-treated conditioned glia were incubated with LPS for 3 h (Fig. 2c). To media (Fig. 4b). When BMP inhibitor was added to the determine if the changes in RNA levels are being astrocyte medium prior to addition of conditioned translated to protein, we determined IFN-γ levels by medium from microglia, statistically significant Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 8 of 18

Fig. 2 BMP7 injection triggers inflammatory changes in the mouse retina. Expression levels of a panel of proinflammatory markers were analyzed by RT-qPCR in RNA samples from mouse retina injected with vehicle or BMP7, harvested 3 and 7 days postinjection (a). At 3 days post-BMP7 injection, about a 2-fold increase in RNA levels, relative to the vehicle controls, was observed in levels of Ifn-γ, Tnf-α,andIl-1β. Seven days post-BMP7 injection, 2-fold increase in levels was observed in Csf, Vegf, Thbs1,andThbs2, and greater than 3-fold increase in Gm-csf, Ifn-γ, Il6, and CD68 RNA levels relative to the vehicle-injected control. Mouse retinal microglial cells treated with BMP7 for 3, 6, 12, and 24 h were also analyzed for changes in RNA levels of inflammatory markers (b), with LPS treatment used as a positive control (c). In vitro treatments showed a significant increase in Ifn-γ levels at the 3-h time point. At 6 h post-BMP7 treatment, mRNA levels of Gm-csf, Ifn-γ, Csf, Tnfα, Il-6,andCd68 were increased to 1.5-fold or greater. By 12 h, we observed no significant differences between BMP7 and vehicle-treated samples. At the 24-h time point, however, we observed significant increases in the levels of Ifn-γ and Thbs. The LPS-treated microglia showed a relative increase in most of the markers, with significant increases observed in levels of Gm-csf, Ifn-γ, Il-6,andThbs2 (c). Protein levels of IFN-γ was also determined via ELISA (d). We observed a 2-fold increase in levels in medium from microglial cells incubated with BMP7 for 24 h and in protein from whole retinal tissue from mice injected with BMP7 for 3 days, when compared to their respective vehicle control. Protein from 7 days BMP7-injected retina showed a 5-fold increase in protein levels compared to the vehicle control. Data shown in graphs represent relative expression levels of RNA or protein of BMP7 or LPS-treated samples to their respective vehicle control. Bars above a level of 1.0 (solid black line) represent an increase in mRNA levels while bars below the level of 1.0 represent a decrease in mRNA levels relative to the corresponding vehicle control. Statistical analysis was performed by unpaired Student’s t test. *p value <0.05. Abbreviations: CD cluster of differentiation, Csf colony stimulating factor, Gm-csf granulocyte macrophage colony stimulating factor, Ifn interferon, Il interleukin, Tnf-α alpha, Thbs thrombospondin, Vegf vascular endothelial growth factor increases were detected in Gfap, Gs, S100-β, Egfr, and toll PLX ablates retinal microglia like receptor-4 (Tlr4; Fig. 4d). Treatment of retinal astro- To further investigate the role of microglia in BMP7- cytes with DMSO or LDN alone or with conditioned mediated gliosis, a means to ablate microglial cells media in presence of DMSO were used as experimental within the retina was sought. Previous reports have controls (Fig. 4a, c). We did not observe any changes shown colony stimulating factor receptor 1 (CSFR1) when cells were treated with LDN alone (Fig. 4a). Treat- inhibitor, PLX3397, to selectively ablate microglia in the ment of retinal astrocytes with conditioned media in the [49]. We have used a variant of the drug, presence of DMSO showed similar changes in expression PLX5622, supplied by Plexxikon Inc. in chow form to as cells treated with conditioned media alone (Fig. 4c). determine its effect on retinal microglia. Starting at Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 9 of 18

the control and PLX treated mice also showed no change (Fig. 6m).

Microglial ablation reduces BMP7-mediated gliosis To determine if microglia were involved in BMP7- mediated gliosis response, mice with ablated microglia (PLX mice) were injected intravitreally with vehicle or BMP7, and mRNA levels of proinflammatory markers or gliosis-related molecules were determined by RT-qPCR 7 days postinjection. As in previous graphs, levels of mRNA are relative to levels in the respective vehicle- treated mice. Mice kept on control chow and treated with BMP7 showed an increase in levels of inflammatory markers including Gm-csf, Ifn-γ, Il-6,andIba1 and gliosis markers including vimentin (Vim), Gfap, Egfr, Mmp9, lipocalin 2 (Lcn2), and thioredoxin interacting protein (Txnip) (Fig. 7a, b). Analysis of inflammatory markers of mice on PLX chow and treated with BMP7 via RT-qPCR also showed only modest increases in levels of Il-1β and Vegf compared to vehicle control (Fig. 7b). mRNA levels Fig. 3 BMP7 alters microglial morphology. Retinal flatmounts from of Gm-csf, Ifn-γ, Il-6, Cd68,andIba1 dropped drastically 1 day BMP7- and vehicle-injected retina were labeled for IBA1 (a, b) when microglia were not present (Fig. 7b). RT-qPCR ana- and analyzed for morphological changes. An increase in the area of lysis of PLX mice 7 days post-BMP7 treatment showed no the microglia was observed when compared to the vehicle control (c). Number of branches and branch length were also assessed for the increase in mRNA levels of gliosis markers compared to treated cells, and increase in the number of branches was observed the PLX vehicle controls (Fig. 7a). Markers indicative of with a decrease in the branch length of the cells incubated with LPS or gliosis were further investigated by examining patterns of BMP7 when compared to the vehicle control (c). Data shown in c represent immunoreactivity for GFAP, S100-β, and neurocan relative change in area and number of branches in BMP7-treated samples (NCAN) (Fig. 8A, B). Three days following vehicle or to the vehicle control. Bars above a level of 1.0 (solid black line) represent an increase while bars below the level of 1.0 represent a BMP7 injection, the PLX mice showed similar levels of decrease in the parameter measured, relative to the corresponding GFAP and S100-β label in control and PLX mice (Fig. 8A vehicle control. Magnification bar in a =50μm, for images (a)and(b). (a, b, d, e, g, h, j, k)). However, 7 days postinjection, the Abbreviation: IBA1 ionized calcium-binding adapter molecule 1 PLX mice showed decreased GFAP and S100-β label in BMP7-injected PLX retinas (Fig. 8B (j, k)), when postnatal day 30, mice were switched to control chow or compared to the control BMP7-injected retinas chow containing 1200 ppm PLX. The mice continued (Fig. 8B (d, e)). NCAN immunofluorescence label did treatment with the inhibitor-laced chow until sacri- not diminish following PLX treatment in comparison ficed 7, 14, or 21 days later. Retinal flatmounts from to controls at either 3 days (Fig. 8A (f, l)) or 7 days control and PLX mice were isolated for 7, 14, and postinjection (Fig. 8B (f, l)). Moreover, levels of 21 days, and labeled for IBA1 and GFAP (Fig. 5). NCAN were increased in vehicle-injected eyes at both Although no apparent change in GFAP was observed 3 and 7 days of PLX-treated mice in comparison to (Fig. 5b–e), there was a clear decrease in the number vehicle-injected eyes of control-treated mice (compare of IBA1+ cells 7 days after starting the PLX diet, and Fig. 8A (c and i) and Fig. 8B (c and i)), supporting a poten- IBA1 immunoreactivity was completely lost by 14 days tial role for microglia in extracellular matrix remodeling. (Fig. 5f–m). Retinal tissue sections from these mice Negative controls showed no label (Additional file 3: were also analyzed for ganglion cells (Brn3a), bipolar Figure S3). Gliosis markers showed similar label in unin- cells (Chx10), Müller glia (SOX9), and horizontal cells jected mice in comparison to the 3 day and 7 day vehicle (Calbindin) (Fig. 6a–d, f–i). Cell counts for labeled injected mice (Additional file 4; Figure S4). Protein levels cells showed no statistically significant change of gliosis markers GFAP, S100-β, and TXNIP were also between the control and PLX-treated mice (Fig. 6k). quantified using western blot (Additional file 5: Figure S5). Retinal flatmounts of PLX and vehicle-treated retinas were also labeled with Gα transducin to label photo- Discussion receptors (Fig. 6e, j). Cell count of labeled images Our lab previously showed that BMP7 is able to trigger showed no statistically significant change in cell reactive gliosis in the retina. Here, we show that the numbers (Fig. 6l). Thickness of retinal sections of Müller cell gliosis triggered by BMP7 is an indirect Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 10 of 18

Fig. 4 Activated microglia secrete factors that trigger retinal gliosis. Conditioned medium from microglial cells incubated with BMP7 or vehicle for 24 h was added to the medium of the retinal astrocytes, directly or pretreated with LDN193189 (c, d). RNA was isolated from these cells 24 h posttreatment and analyzed via RT-qPCR for a panel of gliosis markers. Statistically significant increase in levels of Gfap, Gs, S100-β, Pcan, Egfr, and Tlr4 was observed in astrocytes incubated with conditioned medium added directly or pretreated with LDN193189 (c, d). Cells treated with DMSO (carrier for LDN193189) or LDN only (a) or cells pretreated with DMSO and conditioned medium from BMP7 or vehicle-treated microglia (b) were used as experimental controls. Data shown in graphs represent relative expression levels of RNA in retinal astrocyte cells treated with LDN193189 relative to DMSO (a) or with conditioned media from BMP7-treated microglial cells relative to retinal astrocyte cells treated with conditioned media from vehicle-treated microglia (b–d). Bars above a level of 1.0 (solid black line) represent an increase in mRNA levels while bars below the level of 1.0 represent a decrease in mRNA levels relative to the corresponding vehicle control. Statistical analysis was performed by unpaired Student’s t test. *p value <0.05. Abbreviations: Egfr epidermal growth factor receptor, Gfap glial fibrillary acidic protein, Pcan phosphacan, Tlr toll like receptor effect resulting from microglial activation to a proinflam- [21, 55–57]. The BMP receptors type 1A and 1B matory state. Following exposure to BMP7, microglia up- regulate hypertrophic and scarring responses of regulated at least two molecules, IFN-γ and IL-6, both of astrocytes following spinal cord injury [12]. In the which have been shown in previous studies to trigger glio- retina, BMP signaling components, phospho SMAD sis [50–54]. The CSFR1 inhibitor PLX was used to specif- 1/5/8, have also shown to be upregulated following ically target and ablate retinal microglia without affecting NMDA-induced injury and promote retinal ganglion numbers of other retinal cells, in order to show that the cell’s survival [8]. We observed an increase in BMP7 triggers gliosis through microglial activation. We pTAK1 label in IBA-labeled cells in the retina, as well observed that BMP7 injection into retinas lacking micro- as in other cells of the inner nuclear layer. Increases glia produced an abated inflammatory response and a in expression of pTAK1 in neurons have been previ- complete loss of gliosis, suggesting an important role for ously reported in the brain following cerebral ischemia the microglia in mediating the gliosis response. and is known to be expressed in axonal arbors of sensory neurons [58, 59]. BMPs have also been shown BMP pathway in retinal disease to be important in retinal cell proliferation and regen- BMPs have been previously shown to be regulated in eration in the chick retina [60]. Ueki and Reh [61] injury and disease models of the CNS and retina showed that SMAD upregulation was essential in Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 11 of 18

Fig. 5 PLX ablates microglia in the retina. Mice were fed with chow-containing PLX or vehicle dye to determine ablation of microglia in the retina. a Schematic describing the time points for which the mice were fed with chow-containing PLX, following which eyes were harvested. Retinal flatmounts prepared from the eyes harvested at 7, 14, and 21 days were labeled for GFAP or IBA1 (b–i). Insets in b–e indicate the flatmount outline and from where the images (b–i) was taken. While GFAP did not show any difference between the stages examine (b–e), there was a significant decrease in IBA1 label in mice kept on PLX diet for 7 days (f, g). By 14 days, no IBA1 label was found in the retinal flatmount and this absence persisted into the 21-day time point (h, i). Retinal sections control and PLX-treated mice labeled for IBA1 to show loss of microglia in the deeper layers of the retina (j–m). Statistical analysis was performed by unpaired Student’s t test. *p value <0.05. Magnification bar in b =50μm, for images b–i. Magnification bar in j =50μm, for images j–m. Abbreviations: GFAP glial fibrillary acidic protein, IBA1 ionized calcium-binding adapter molecule 1 mediating EGF dependent Müller glial cell prolifera- response. Microglia are the resident macrophages in the tion in the mouse. The presence of BMPs in disease retina. Similar to the macroglia, these cells also undergo states is consistent with a potential role for them activation. Their activation has been observed in various playing a role in retinal gliosis. disease and injury models, such as retinitis pigmentosa, diabetic retinopathy, retinal detachment, and glaucoma Activated microglia drive retinal gliosis [62–66]. Activated microglia change morphology from a We had previously reported that BMP7 was able to trigger ramified cell to an amoeboid cell, along with changes in gliosis in retinal glia in vitro and in vivo. However, we expression of cell surface markers, such as the cluster of observed a higher response in the in vivo model, which differentiation molecule 11b (CD11b), CD68, major suggested there may be other cells involved in this histocompatibility complexes (MHC), scavenger receptors, Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 12 of 18

Fig. 6 PLX ablates microglia without affecting other retinal cells. Retinal sections from mice kept on the PLX or vehicle chow diet for 14 days were labeled for ganglion cells (Brn3a; a, f), bipolar cells (Chx10; b, g), Müller glia (Sox9; c, h), and horizontal cells (Calbindin; d, i). Cell counts of images for the labeled markers (n = 8) showed no difference between the control and PLX-treated mice (k; Images taken were within 200 μm from the optic nerve). 60× images of retinal flatmounts of PLX and control-treated mice were labeled for Gα transducin (e, j). Cell counts showed no difference in the two treatments (l). Retinal thickness was also assessed in control and PLX retinas and showed no difference (m). Magnification bar in a =50μm, for images (a–d, f–i). Magnification bar in e =10μm, for images (e, j)

and TLR and secreted actors such as RANTES, interferon, retinal detachment and retinitis pigmentosa [68–70]. How- interleukins, and TNFα. These changes serve to en- ever, research has also revealed that there are differences in hance the phagocytic effect of the microglial cells as the response of the two cell types. GFAP upregulation was well as the cytotoxic effect on injured cells and for- observed in Müller glia and not in the astrocytes in rats eign pathogen [23, 67]. Müller glia also undergo acti- subjected to episcleral vein cauterization [71]. Similarly, up- vation following disruption of retinal homeostasis. regulation of GFAP was observed in the Müller glial cells in The reactive Müller glia hypertrophy and upregulated the retina subjected to laser-induced ocular hypertension, expression of various growth factors, reactive oxygen while the astrocytes of the contralateral control eyes also species scavengers, protect neurons from excitotoxi- exhibited an increase in GFAP and a change in the area city and, in some organisms, can regenerate retinal covered by the astrocytes [72, 73]. The differences observed neurons. These changes serve to protect the damaged may suggest distinct functional roles for the astrocytes and retina. However, gliosis can also have detrimental Müller glia, which cooperate to restore retinal homeostasis. effects by remodeling the extra cellular matrix and Here, we observed a decreased gliosis response in the due to loss of normal glial functions which are neces- retina following BMP7 treatment in mice lacking micro- sary for normal neuronal activity [6, 7]. glia. We used a novel CSF1R inhibitor (PLX) to select- The retinal astrocytes and Müller glial cells exhibit simi- ively ablate microglia. Following microglial ablation, lar responses to injury, such as hypertrophy, upregulation mice were treated with BMP7 to assess gliosis in the of GFAP, vimentin, and GS, as observed in rat models of retina. The inclusion of the inhibitor in the chow Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 13 of 18

Fig. 7 Effect of BMP7 is diminished in the absence of microglia—RNA levels. BMP7 or vehicle was injected intravitreally into the eyes of mice kept on regular chow or PLX chow and harvested 7 days postinjection. RNA isolated from the retina were analyzed via RT-qPCR for changes in levels of inflammatory markers (a) and gliosis markers (b). Mice kept on the control chow and injected with BMP7 showed a relative increase by 2-fold or greater of inflammatory markers: Gm-csf, Csf, Ifn-γ, Il-6, Vegf, Thbs1, Thbs2, and CD68 (b). Gliosis markers Gfap, Vim, S100-β, Gs, Ncan, Mmp9, Lcn2, and Txnip showed a 2-fold increase or more in these mice (a). Data shown in graphs (a, b) represent relative expression levels of RNA in mouse retina to the respective vehicle treatments. Bars above a level of 1.0 (solid black line) represent an increase in mRNA levels while bars below the level of 1.0 represent a decrease in mRNA levels relative to the corresponding vehicle control. Mice kept on the PLX chow and injected with BMP7 showed a 2-fold increase in inflammatory markers Il-1β and Vegf, while all the gliosis markers showed relatively unchanged RNA levels (a, b). Statistical analysis was performed by one way ANOVA with post hoc Tukey test. Significant difference from vehicle-injected control mice *p value <0.05. Significant difference from BMP7 injected control mice #p value <0.05. Abbreviations: Csf colony stimulating factor, Egfr epidermal growth factor receptor, Gm-csf granulocyte macrophage colony stimulating factor, Gfap glial fibrillary acidic protein, Gs glutamine synthetase, Ifn-γ interferon-gamma, Il interleukin, Lcn lipocalin, Mmp matrix metalloproteinase, Ncan neurocan, Pcan phosphacan, Txnip thioredoxin interacting protein, Tnf-α tumor necrosis factor alpha, Thbs thrombospondin, Tlr toll like receptor, Vim vimentin, Vegf vascular endothelial growth factor

allowed its continual application over a longer period of BMP and inflammation time, enabling the maintenance of a microglia-free Activation of microglia and macroglia have been studied environment in the retina in which we could test the in various models. While there are differences in the role of the microglia in BMP7-mediated gliosis. With- responses of the two glial populations, they do exhibit out continual application of the inhibitor, microglia similarities. These include regulation of inflammatory could repopulate the retina from one of two sources; markers, antigen presentation complexes, and various bone marrow-derived stem cells can penetrate the factors such as IFN, TNFα, and TLR [3, 6, 36]. While blood-brain barrier and differentiate into microglia or several different factors have been shown to regulate residential microglia can proliferate and replace lost macrophage and microglia activation, the effect of BMPs cells [74, 75]. The two sources of microglia are not is still not completely characterized [39, 40, 76, 77]. equivalent; residential microglia primarily give rise to BMP6 regulates expression of inflammatory markers microglia that display an M1 inflammatory phenotype, such as IL-6, IL-1β, and nitric oxide synthase in macro- whereas the bone marrow-derived cells give rise to phages [78–80]. In addition, more recent studies indicate microglia with an M2 anti-inflammatory phenotype that BMP exposure particularly leads to the M2 or anti-- [75]. At any rate, in order for us to test the role of inflammatory phenotype of the macrophages promoting BMP7 in indirectly triggering gliosis, we had to main- tissue repair [81–84]. Microglia are descendants of im- tain a microglial-free environment for the duration of mature macrophages and are thought to act as macro- the experiments. phages in disease and injury states [85]. In our studies, Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 14 of 18

Fig. 8 Effect of BMP7 on gliosis in absence of microglia—localization of gliosis markers. Mouse retinal sections from eyes injected with vehicle or BMP7 were labeled for gliosis markers GFAP (A, B (a, d, g, j)), S100-β (A, B (b, e, h, k)), and NCAN (A, B (c, f, j, l)). Mice kept on the PLX diet did not show an increase in label for the gliosis markers GFAP and S100-β BMP7 or vehicle-injected retina 3 and 7 days postinjection (A, B (g, h, j, k)). NCAN label appeared to be similar in the BMP7 injected and the respective age-matched vehicle controls in mice kept on the PLX chow (A, B (i, l)). Mice kept on the control chow and injected with BMP7 clearly showed an increase in GFAP, S100-β, and NCAN levels 7 days postinjection, when compared to their respective vehicle control (B (a–f)). Three days postinjection, there is an increase in GFAP and NCAN label in BMP7-injected retinas in comparison to the respective vehicle control-injected retinas (A (a, c, d, f)). When comparing mice kept on control chow or the PLX chow, there is an increase in GFAP and S100-β label in the mice kept on control chow in comparison to the mice kept on the PLX chow, 7 days post BMP7 injection (B (d, e, j, k)). GFAP and S100-β label in mice kept on control chow and PLX chow appears to be similar in the BMP7-injected retinas, 3 days postinjection (A (d, e, j, k)). Magnification bar in (B,(c)) = 50 μm, for images A, B (a–l)

BMP7 increased the proinflammatory state of the micro- In the PLX-treated mice (both vehicle and BMP7- glia. Further studies are necessary to determine if all injected), we also observe an increase in neurocan levels microglia respond to BMP7 by increasing proinflamma- in the retina. Müller glia secrete MMPs that regulate tory markers or if this is a response unique to certain neurocan levels in the extracellular matrix. In addition, populations of microglia. microglia also secrete these enzymes [86, 87]. Their In this study, we observed that microglia showed an upregulation has been observed in the CNS during upregulation of inflammatory markers in response to inflammation in various injury models. Furthermore, BMP7 treatment, indicative of activation. Furthermore, microglia-derived factors such as TNF-α have also shown in the PLX treated mice, the gliosis response was to regulate MMP expression by the Müller glia [88]. Thus, subdued in comparison to control BMP7 treated retinas, we propose that the lack of microglia in the retina contrib- suggesting that microglia are an essential mediator of utes to the increase in neurocan by regulating MMP levels retinal gliosis. These results support our hypothesis that either directly or indirectly by regulating Müller glia. microglia are activated by BMP7, which in turn regulate Comparing the mRNA and protein levels in the factors causing Müller cell gliosis. control and PLX-injected retinas, we observed a difference Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 15 of 18

in expression patterns (Fig. 7, Additional file 5: Figure S5). [23, 100]. Morphological changes and increases in RNA Although the mRNA levels of S100 and TXNIP was levels of inflammatory markers in the microglia following reduced in the BMP7-injected PLX mice, we did not BMP7 treatment indicate activation of the microglia. We observe a similar change at the protein level. Non- observed in our analysis that RNA levels of Ifn-γ, Il-6,Vegf, correlation between mRNA and protein levels has and Thbs1 to be greater following Müller glia activation. been noted in other studies [89–92]. mRNA transla- Previously, Cotinet et.al and Goureau showed that IFN-γ tion and protein stability in the cell is regulated by can trigger Müller glia to regulate TNF-α and nitric oxide multiple systems including micro RNAs (miRNAs), (NO) [101, 102]. Similarly, IL-6 has been shown to induce mRNA localization translational repression, and pro- Müller glia-derived progenitor cells in the injured zebra- tein stability [92, 93]. miRNAs have been previously fish and chick retina [40, 103]. We propose that BMP7 reported to be regulated in neural tissue under causes activation of microglia, which leads to upregulation conditions of stress [94–96]. Furthermore, BMPs can of factors such as IFN-γ and IL-6, which in turn trigger regulate translation by regulating cytoplasmic polyade- Müller cell gliosis. nylation element binding protein (CPEB) via the TAK Our findings indicated an important role for microglia pathway [97, 98]. Further studies will be required to in Müller cell gliosis in the murine retina. However, the determine what pathway(s) mediate this non- mechanism and potential factors that play a role in correlation between the mRNA and protein levels. microglia and Müller glia interactions are not known. Future studies will aim to identify the potential role of IFN-γ and IL-6, upregulated by BMP7 in the retina, in Microglia release inflammatory factors prior to formation microglia function, and gliosis. of gliosis We observed a decrease in expression of GFAP and S100- Conclusions β in mice kept on the PLX diet and treated with BMP7. Our findings indicate that retinal microglia are essential BMP7 treatment also revealed decreased RNA levels of in regulating retinal gliosis. The expression of down- gliosis and inflammatory markers in PLX mice when com- stream BMP signaling components in the retinal pared to the mice kept on the normal diet. Previously, it microglia, as well as the decrease in retinal gliosis in has been reported that microglia respond early to changes PLX5622-treated mice demonstrate that BMP7 can in microenvironment and become activated. Bosco et al. regulate gliosis indirectly by activating the retinal micro- showed that microglia become activated early in the ret- glia. Additionally, we show that regulation of retinal ina, prior to any increases in IOP in the DBA/2J mice gliosis by microglia could be mediated by IFN-γ or IL-6. [25]. Similarly, early activation of microglia has also been Further studies will help evaluate the role of these observed and implicated in progression of Parkinson’sdis- factors in this response. ease [99]. Furthermore, in the ocular hypertension mouse model studied in Gallego et al., the authors suggest that upregulation of MHC-II in microglia in the controlateral Additional files eye regulated the morphological changes of retinal astro- Additional file 1: Figure S1. PU.1 localizes with retinal microglia. cytes [73]. Thus, we propose that microglia respond to the Co-label of PU.1 antibody with antibody against GFP that cross-reacts BMP7 first and become activated. These activated micro- with YFP in a retinal section from P30 mice which have YFP tag on glia upregulate factors, which in turn can trigger Müller vascular endothelial cadherin (VE-YFP), a marker expressed in endothelial cells (A-D). No co-label of PU.1 was observed with VE-YFP. PU.1 was also cell gliosis. Consistent with this notion, our findings indi- co-labeled with microglia marker IBA1 to show localization was restricted to cate the Ifn-γ and other inflammatory factors were microglial cells (E-H). Hoechst merged with the images of green and red upregulated as early as 3 h following incubation of micro- channels are shown in D and H. Magnification bar in E = 50 μm, for images A–H. (TIF 857 kb) glial cells with BMP7 in vitro, and these levels were fur- Additional file 2: Figure S2. Expression of BMP signaling molecules in ther increased 6, 12, and 24 h postincubation with BMP7. microglia in vehicle and BMP7-injected retinas. Retinal sections from In contrast, factors associated with gliosis do not begin to P30 mouse injected with vehicle or BMP7 24 h postinjection were increase until 3 days in vivo, with most markers increasing double-labeled with antibodies that labels microglia cytoplasm (IBA1) and phospho SMAD 1/5/9 (pSMAD; A–F) or phospho TAK1 (pTAK1; G–L). Thin after 7 days. plane confocal microscopy images with y,z (strips to right of the panel) and x,z planes (strips at the bottom of the panels) shown in C, F, I and L. pSMAD-labeled cells were primarily found in the GCL in the vehicle-treated Potential factors regulating microglia-mediated activation retina, with some co-localization with the cytoplasmic microglial marker IBA1 of Müller glia (A-C). The BMP7-injected retina had an increase in pSMAD expression in the INL as well as substantial co-localization with IBA1 (D–F). Vehicle-injected retina Previous studies looking into microglia and macroglia showed pTAK1 expression in the GCL with little to no IBA1 co-localization interactions have revealed several secreted as well as (G–I), while the BMP7-injected retinas showed increased levels of pTAK1 levels membrane bound factors which could activate the in the INL, as well as significant co-localization with IBA1 (J–L). Magnification bar in A = 50 μm, for images A–L. (TIF 688 kb) macroglia, such as IL-1β,IL-18,TGF-β,andTNF-α Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 16 of 18

Additional file 3: Figure S3. Negative control of immunofluorescence Ethics approval labels. Retinal sections from P30 mouse labeled with rabbit immunoglobulin All procedures were in accordance with the guidelines set by the Institutional G(RbtIgG;A–C, D, F), mouse IgG (Mse IgG; E, F), and sheep IgG (G, H) to Animal Care and Use Committee (IACUC) at the school of science IUPUI determine background fluorescence. Images of sections labeled with the (protocol number SC230R). nuclear stain, Hoechst merged with the images of green and red channels are shown in C and F. Panels A–C represent sections, which were labeled with IgG following the procedure used for tyramide amplification when Publisher’sNote using two antibodies for the same species. Images in D–Frepresentsections Springer Nature remains neutral with regard to jurisdictional claims in published co-labeled with rabbit and mouse IgG. Images A–C are negative controls for maps and institutional affiliations. Fig. 1 and Additional file 1: Figure S1. Images D–F are negative controls for sections labeled with GFAP, S100-β,Calbindin,Brn3a,Chx10,Sox9,andIBA1. Author details 1 Images G and H are negative control sections for NCAN-labeled slides. Department of Biology, Indiana University-Purdue University Indianapolis, 2 Magnification bar in A = 50 μm, for images A–H. (TIF 465 kb) 723 W Michigan St, SL306, Indianapolis, IN 46202, USA. Center for Developmental and Regenerative Biology, Indiana University-Purdue Additional file 4: Figure S4. IF label of retinas for GFAP, S-100-β, and University Indianapolis, 723 W Michigan St, Indianapolis, IN 46202, USA. NCAN in P30 uninjected and 3 and 7 days vehicle-injected retinas. Retinal 3Department of Ophthalmology and Visual Sciences, University of Wisconsin sections from uninjected P30 mouse, vehicle-injected P30 mouse, obtained School of Medicine and Public Health, 1111 Highland Avenue, 9453 WIMR, 3 and 7 days postinjection, labeled for GFAP (A, D, G), S100-β (B, E, H), and Madison, WI 53705, USA. 4Department of Pediatrics, University of Wisconsin NCAN (C, F, I). Label for all three markers appears to be similar in the School of Medicine and Public Health, 1111 Highland Avenue, 9453 WIMR, uninjected and the vehicle-injected retinas. Magnification bar in A = 50 μm, Madison, WI 53705, USA. for images A–I. (TIF 5611 kb) Additional file 5: Figure S5. Protein levels in PLX-treated mice. Protein Received: 2 December 2016 Accepted: 27 March 2017 isolated from control and PLX-treated mice injected with vehicle or BMP7 changes in protein levels of gliosis markers GFAP, S100-β, and TXNIP, with β-Tubulin used as a loading control. GFAP showed elevated levels in the BMP7-injected control mice, while PLX mice had GFAP levels similar to the References vehicle injection. S100-β was elevated in the 3 and 7 days BMP7-injected 1. Jadhav AP, Roesch K, Cepko CL. Development and neurogenic potential of – PLX mice as well as in the 7 days BMP7-injected control mice, compared to Muller glial cells in the vertebrate retina. Prog Retin Eye Res. 2009;28(4):249 62. the respective vehicle controls. TXNIP levels did not change in the control 2. Watanabe T, Raff MC. Retinal astrocytes are immigrants from the optic – and PLX mice injected with vehicle or BMP7 3 days postinjection. Seven nerve. Nature. 1988;332(6167):834 7. days postinjection, TXNIP levels did increase in the control BMP-injected 3. Li L, Eter N, Heiduschka P. The microglia in healthy and diseased retina. Exp – mice, while no such change was observed in the PLX mice. No statistical Eye Res. 2015;136:116 30. significance was observed in the densitometric analysis (B) of blots from 4. Seo JH, et al. Oligodendroglia in the avian retina: immunocytochemical – (A). (TIF 472 kb) demonstration in the adult bird. J Neurosci Res. 2001;65(2):173 83. 5. Fischer AJ, et al. A novel type of glial cell in the retina is stimulated by insulin-like growth factor 1 and may exacerbate damage to neurons and Abbreviations Muller glia. Glia. 2010;58(6):633–49. BMP: Bone morphogenetic protein; CCL: Chemokine ligand; CD: Cluster of 6. Reichenbach A, Bringmann A. Muller cells in the healthy and diseased differentiation; CSF: Colony stimulating factor; EGFR: Epidermal growth factor retina, Muller Cells in the Healthy and Diseased Retina. 2010. p. 1–417. receptor; GFAP: Glial fibrillary acidic protein; GM-CSF: Granulocyte macrophage 7. Bringmann A, et al. Cellular signaling and factors involved in Muller cell colony stimulating factor; GS: Glutamine synthetase; IFN: Interferon; IL: Interleukin; gliosis: neuroprotective and detrimental effects. Prog Retin Eye Res. 2009; LCN: Lipocalin; MMP: Matrix metalloproteinases; NCAN: Neurocan; 28(6):423–51. PCAN: Phosphacan; TAK: TGF-β activated kinase; THBS: Thrombospondin; 8. Ueki Y, Reh TA. Activation of BMP-Smad1/5/8 signaling promotes survival of TLR: Toll like receptor; TNF: Tumor necrosis factor; TXNIP: Thioredoxin interacting retinal ganglion cells after damage in vivo. PLoS ONE. 2012;7(6):e38690. protein; VEGF: Vascular endothelial growth factor; VIM: Vimentin 9. Lutty GA. Effects of diabetes on the eye. Invest Ophthalmol Vis Sci. 2013;54(14):ORSF81–7. Acknowledgements 10. Pekny M, Wilhelmsson U, Pekna M. The dual role of astrocyte activation and The authors would like to thank Plexxikon Inc. for chow laced with PLX5622. reactive gliosis. Neurosci Lett. 2014;565:30–8. 11. Dharmarajan S, et al. Bone morphogenetic protein 7 regulates reactive Funding gliosis in retinal astrocytes and Muller glia. Mol Vis. 2014;20:1085–108. TBA is supported by RO1EY019525, CMS is supported by R21 EY023024, and 12. Sahni V, et al. BMPR1a and BMPR1b signaling exert opposing effects on NS is supported by R24 EY022883 and P30 EY016665 from the National gliosis after spinal cord injury. J Neurosci. 2010;30(5):1839–55. Institutes of Health and an unrestricted departmental award from Research 13. Goldman D. Muller glial cell reprogramming and retina regeneration. Nat to Prevent Blindness. Rev Neurosci. 2014;15(7):431–42. 14. Martinez G, et al. Expression of bone morphogenetic protein-6 and Availability of data and materials transforming growth factor-beta1 in the rat brain after a mild and reversible The datasets during and/or analyzed during the current study available from ischemic damage. Brain Res. 2001;894(1):1–11. the corresponding author on reasonable request. 15. Bragdon B, et al. Bone morphogenetic proteins: a critical review. Cell Signal. 2011;23(4):609–20. Authors’ contributions 16. Miyazono K, Kamiya Y, Morikawa M. Bone morphogenetic protein receptors SD performed in vitro and in vivo treatments. TBA and SD analyzed and and signal transduction. J Biochem. 2010;147(1):35–51. interpreted the in vitro and in vivo data. CM managed the mouse colonies 17. Fuller ML, et al. Bone morphogenetic proteins promote gliosis in from which the microglia and retinal astrocytes were isolated. DF isolated demyelinating spinal cord lesions. Ann Neurol. 2007;62(3):288–300. the mouse microglial cells. NS helped with the microglial and astrocyte 18. Luan LJ, et al. Post-hypoxic and ischemic of BMP-7 in the isolations. TBA, SD, and NS contributed in writing the manuscript. All cerebral cortex and caudate-putamen tissue of rat. Acta Histochem. authors read and approved the final manuscript. 2015;117(2):148–54. 19. Wordinger RJ, et al. Effects of TGF-beta2, BMP-4, and gremlin in the Competing interests trabecular meshwork: implications for glaucoma. Invest Ophthalmol Vis Sci. The authors declare that they have no competing interests. 2007;48(3):1191–200. 20. Woiciechowsky C, et al. Brain-IL-1 beta triggers through Consent for publication induction of IL-6: inhibition by propranolol and IL-10. Med Sci Monit. Not applicable. 2004;10(9):Br325–30. Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 17 of 18

21. Hussein KA, et al. Bone morphogenetic protein 2: a potential new player in 52. Corbin JG, et al. Targeted CNS expression of interferon-gamma in the pathogenesis of diabetic retinopathy. Exp Eye Res. 2014;125:79–88. transgenic mice leads to hypomyelination, reactive gliosis, and abnormal 22. Santos AM, et al. Embryonic and postnatal development of microglial cells cerebellar development. Mol Cell Neurosci. 1996;7(5):354–70. in the mouse retina. J Comp Neurol. 2008;506(2):224–39. 53. Chakrabarty P, et al. Massive gliosis induced by interleukin-6 suppresses 23. Langmann T. Microglia activation in retinal degeneration. J Leukoc Biol. Abeta deposition in vivo: evidence against inflammation as a driving force 2007;81(6):1345–51. for deposition. Faseb J. 2010;24(2):548–59. 24. Chen L, Yang P, Kijlstra A. Distribution, markers, and functions of retinal 54. Chiang CS, et al. Reactive gliosis as a consequence of interleukin-6 expression microglia. Ocul Immunol Inflamm. 2002;10(1):27–39. in the brain—studies in transgenic mice. Dev Neurosci. 1994;16(3-4):212–21. 25. Bosco A, Steele MR, Vetter ML. Early microglia activation in a mouse model 55. Setoguchi T, et al. Traumatic injury-induced BMP7 expression in the adult of chronic glaucoma. J Comp Neurol. 2011;519(4):599–620. rat spinal cord. Brain Res. 2001;921(1-2):219–25. 26. Wang MH, et al. Macroglia-microglia interactions via TSPO signaling regulates 56. Matsuura I, et al. BMP inhibition enhances axonal growth and functional microglial activation in the mouse retina. J Neurosci. 2014;34(10):3793–806. recovery after spinal cord injury. J Neurochem. 2008;105(4):1471–9. 27. Karlstetter M, et al. Retinal microglia: just bystander or target for therapy? 57. Hollborn M, et al. Changes in retinal gene expression in proliferative Prog Retin Eye Res. 2015;45:30–57. vitreoretinopathy: glial cell expression of HB-EGF. Mol Vis. 2005;11:397–413. 28. Boche D, Perry VH, Nicoll JAR. Review: activation patterns of microglia and 58. Lilley BN, Pan YA, Sanes JR. SAD kinases sculpt axonal arbors of sensory their identification in the human brain. Neuropathol Appl Neurobiol. neurons through long- and short-term responses to neurotrophin signals. 2013;39(1):3–18. Neuron. 2013;79(1):39–53. 29. Crain JM, Nikodemova M, Watters JJ. Microglia express distinct M1 and M2 59. Neubert M, et al. Acute inhibition of TAK1 protects against neuronal death phenotypic markers in the postnatal and adult in in cerebral ischemia. Cell Death Differ. 2011;18(9):1521–30. male and female mice. J Neurosci Res. 2013;91(9):1143–51. 60. Haynes T, et al. BMP signaling mediates stem/progenitor cell-induced retina 30. Chhor V, et al. Characterization of phenotype markers and neuronotoxic potential regeneration. Proc Natl Acad Sci U S A. 2007;104(51):20380–5. of polarised primary microglia in vitro. Brain Behav Immun. 2013;32:70–85. 61. Ueki Y, Reh TA. EGF stimulates Muller glial proliferation via a BMP- 31. Jaguin M, et al. Polarization profiles of human M-CSF-generated macrophages dependent mechanism. Glia. 2013;61(5):778–89. and comparison of M1-markers in classically activated macrophages from 62. Yoshida N, et al. Laboratory evidence of sustained chronic inflammatory GM-CSF and M-CSF origin. Cell Immunol. 2013;281(1):51–61. reaction in retinitis pigmentosa. Ophthalmology. 2013;120(1):E5–12. 32. Crane MJ, et al. The monocyte to macrophage transition in the murine 63. Grigsby JG, et al. The role of microglia in diabetic retinopathy. J sterile wound. PLoS ONE. 2014;9(1):e86660. Ophthalmol. 2014. Article ID 705783. http://dx.doi.org/10.1155/2014/ 33. Harry GJ. Microglia during development and aging. Pharmacol Ther. 705783. 2013;139(3):313–26. 64. Wang JW, et al. Retinal microglia in glaucoma. J Glaucoma. 2016;25(5):459–65. 34. Zeng HY, et al. Identification of sequential events and factors associated 65. Fischer AJ, Zelinka C, Milani-Nejad N. Reactive retinal microglia, with microglial activation, migration, and cytotoxicity in retinal neuronal survival, and the formation of retinal folds and detachments. degeneration in rd mice. Invest Ophthalmol Vis Sci. 2005;46(8):2992–9. Glia. 2015;63(2):313–27. 35. Zeng XX, Ng YK, Ling EA. Neuronal and microglial response in the retina of 66. Karlstetter M, Ebert S, Langmann T. Microglia in the healthy and streptozotocin-induced diabetic rats. Vis Neurosci. 2000;17(3):463–71. degenerating retina: insights from novel mouse models. Immunobiology. 36. Kumar A, Shamsuddin N. Retinal Muller glia initiate innate response to 2010;215(9-10):685–91. infectious stimuli via toll-like receptor signaling. PLoS ONE. 2012;7(1):e29830. 67. Luo XG, Chen SD. The changing phenotype of microglia from homeostasis 37. Xue W, et al. Ciliary neurotrophic factor induces genes associated with to disease. Transl Neurodegener. 2012;1(1):9. inflammation and gliosis in the retina: a gene profiling study of flow-sorted, 68. Luna G, et al. Expression profiles of nestin and synemin in reactive Muller cells. PLoS ONE. 2011;6(5):e20326. astrocytes and Muller cells following retinal injury: a comparison with glial 38. Balasingam V, Yong VW. Attenuation of astroglial reactivity by interleukin-10. fibrillar acidic protein and vimentin. Mol Vis. 2010;16:2511–23. J Neurosci. 1996;16(9):2945–55. 69. Chang ML, et al. Reactive changes of retinal astrocytes and Muller glial cells 39. Wang M, et al. Adaptive Muller cell responses to microglial activation in kainate-induced neuroexcitotoxicity. J Anat. 2007;210(1):54–65. mediate neuroprotection and coordinate inflammation in the retina. J 70. Fernandez-Sanchez L, et al. Astrocytes and Muller cell alterations during Neuroinflammation. 2011;8:173. retinal degeneration in a transgenic rat model of retinitis pigmentosa. Front 40. Fischer AJ, et al. Reactive microglia and macrophage facilitate the formation Cell Neurosci. 2015;9:484. of Muller glia-derived retinal progenitors. Glia. 2014;62(10):1608–28. 71. Kanamori A, et al. Long-term glial reactivity in rat retinas ipsilateral and 41. Scheef E, et al. Isolation and characterization of murine retinal astrocytes. contralateral to experimental glaucoma. Exp Eye Res. 2005;81(1):48–56. Mol Vis. 2005;11:613–24. 72. Ramirez AI, et al. Quantification of the effect of different levels of IOP in the 42. Roque RS, Caldwell RB. Isolation and culture of retinal microglia. Curr Eye astroglia of the rat retina ipsilateral and contralateral to experimental Res. 1993;12(3):285–90. glaucoma. Invest Ophthalmol Vis Sci. 2010;51(11):5690–6. 43. Cuny GD, et al. Structure-activity relationship study of bone morphogenetic 73. Gallego BI, et al. IOP induces upregulation of GFAP and MHC-II and protein (BMP) signaling inhibitors. Bioorg Med Chem Lett. 2008;18(15):4388–92. microglia reactivity in mice retina contralateral to experimental glaucoma. 44. Srinivas S, et al. Cre reporter strains produced by targeted insertion of EYFP J Neuroinflammation. 2012;9:92. and ECFP into the ROSA26 locus. BMC Dev Biol. 2001;1:4. 74. Elmore MR, et al. Characterizing newly repopulated microglia in the adult 45. Alva JA, et al. VE-Cadherin-Cre-recombinase transgenic mouse: a tool for lineage mouse: impacts on animal behavior, cell morphology, and analysis and gene deletion in endothelial cells. Dev Dyn. 2006;235(3):759–67. neuroinflammation. PLoS ONE. 2015;10(4):e0122912. 46. Tual-Chalot S, et al. Whole mount immunofluorescent staining of the neonatal 75. Jin N. et al. Friend or foe? Resident microglia vs bone marrow-derived mouse retina to investigate angiogenesis in vivo. J Vis Exp. 2013;77:e50546. microglia and their roles in the retinal degeneration. Mol. Neurobiol. 2016:1-19. 47. Ferreira TA, et al. Neuronal morphometry directly from bitmap images. Nat doi:10.1007/s12035-016-9960-9. Methods. 2014;11(10):982–4. 76. Harada T, et al. Microglia-Muller glia cell interactions control neurotrophic 48. Kreutzberg GW. Microglia: a sensor for pathological events in the CNS. factor production during light-induced retinal degeneration. J Neurosci. Trends Neurosci. 1996;19(8):312–8. 2002;22(21):9228–36. 49. Elmore MRP, et al. Colony-stimulating factor 1 receptor signaling is 77. Wang M, Wong WT. Microglia-Muller cell interactions in the retina. Adv Exp necessary for microglia viability, unmasking a microglia progenitor cell in Med Biol. 2014;801:333–8. the adult brain. Neuron. 2014;82(2):380–97. 78. Lee GT, et al. Induction of interleukin-6 expression by bone morphogenetic 50. Lee SC, et al. Induction of nitric oxide synthase activity in human protein-6 in macrophages requires both SMAD and p38 signaling pathways. astrocytes by interleukin-1 beta and interferon-gamma. J Neuroimmunol. J Biol Chem. 2010;285(50):39401–8. 1993;46(1-2):19–24. 79. Hong JH, et al. Effect of bone morphogenetic protein-6 on macrophages. 51. Yong VW, et al. Gamma-interferon promotes proliferation of adult human Immunology. 2009;128(1):e442–50. astrocytes in vitro and reactive gliosis in the adult mouse brain in vivo. Proc 80. Kwon SJ, et al. Bone morphogenetic protein-6 induces the expression of inducible Natl Acad Sci U S A. 1991;88(16):7016–20. nitric oxide synthase in macrophages. Immunology. 2009;128(1):e758–65. Dharmarajan et al. Journal of Neuroinflammation (2017) 14:76 Page 18 of 18

81. Singla DK, Singla R, Wang J. BMP-7 treatment increases M2 macrophage differentiation and reduces inflammation and plaque formation in apo E-/- mice. PLoS ONE. 2016;11(1):e0147897. 82. Rocher C, Singla DK. SMAD-PI3K-Akt-mTOR pathway mediates BMP-7 polarization of monocytes into M2 macrophages. PLoS ONE. 2013;8(12):e84009. 83. Urbina P, Singla DK. BMP-7 attenuates adverse cardiac remodeling mediated through M2 macrophages in prediabetic cardiomyopathy. Am J Phys Heart Circ Phys. 2014;307(5):H762–72. 84. Rocher C, et al. Bone morphogenetic protein 7 polarizes THP-1 cells into M2 macrophages. Can J Physiol Pharmacol. 2012;90(7):947–51. 85. Wake H, Moorhouse AJ, Nabekura J. Functions of microglia in the central nervous system—beyond the immune response. Neuron Glia Biol. 2011;7(1):47–53. 86. Nuttall RK, et al. Metalloproteinases are enriched in microglia compared with leukocytes and they regulate levels in activated microglia. Glia. 2007;55(5):516–26. 87. del Zoppo GJ, et al. Microglial activation and matrix protease generation during focal cerebral ischemia. . 2007;38(2 Suppl):646–51. 88. Limb GA, et al. Differential expression of matrix metalloproteinases 2 and 9 by glial Muller cells: response to soluble and extracellular matrix-bound tumor necrosis factor-alpha. Am J Pathol. 2002;160(5):1847–55. 89. Koussounadis A, et al. Relationship between differentially expressed mRNA and mRNA-protein correlations in a xenograft model system. Sci Rep. 2015;5:10775. 90. Inman DM, Horner PJ. Reactive nonproliferative gliosis predominates in a chronic mouse model of glaucoma. Glia. 2007;55(9):942–53. 91. Wong RW, Hagen T. Mechanistic target of rapamycin (mTOR) dependent regulation of thioredoxin interacting protein (TXNIP) transcription in hypoxia. Biochem Biophys Res Commun. 2013;433(1):40–6. 92. Maier T, Guell M, Serrano L. Correlation of mRNA and protein in complex biological samples. FEBS Lett. 2009;583(24):3966–73. 93. Di Liegro CM, Schiera G, Di Liegro I. Regulation of mRNA transport, localization and translation in the nervous system of mammals (Review). Int J Mol Med. 2014;33(4):747–62. 94. Liu L, et al. dysregulates microRNAs to modulate cell signaling in rat hippocampus. PLoS ONE. 2014;9(8):e103948. 95. Rajaram K, et al. Dynamic miRNA expression patterns during retinal regeneration in zebrafish: reduced dicer or miRNA expression suppresses proliferation of Muller glia-derived neuronal progenitor cells. Dev Dyn. 2014;243(12):1591–605. 96. Bhalala OG, Srikanth M, Kessler JA. The emerging roles of microRNAs in CNS injuries. Nat Rev Neurol. 2013;9(6):328–39. 97. Kim KC, Hyun Joo S, Shin CY. CPEB1 modulates lipopolysaccharide- mediated iNOS induction in rat primary astrocytes. Biochem Biophys Res Commun. 2011;409(4):687–92. 98. Cargnello M, Roux PP. Activation and function of the MAPKs and their substrates, the MAPK-activated protein kinases. Microbiol Mol Biol Rev. 2011;75(1):50–83. 99. Ouchi Y, et al. Microglial activation and dopamine terminal loss in early Parkinson’s disease. Ann Neurol. 2005;57(2):168–75. 100. Miyoshi K, et al. Interleukin-18-mediated microglia/astrocyte interaction in the spinal cord enhances neuropathic pain processing after nerve injury. J Neurosci. 2008;28(48):12775–87. 101. Goureau O, et al. Induction and regulation of nitric oxide synthase in retinal Muller glial cells. J Neurochem. 1994;63(1):310–7. 102. Cotinet A, et al. Tumor necrosis factor and nitric oxide production by retinal Muller glial cells from rats exhibiting inherited retinal dystrophy. Glia. 1997;20(1):59–69. 103. Zhao XF, et al. Leptin and IL-6 family cytokines synergize to stimulate Muller Submit your next manuscript to BioMed Central – glia reprogramming and retina regeneration. Cell Rep. 2014;9(1):272 84. and we will help you at every step:

• We accept pre-submission inquiries • Our selector tool helps you to find the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript at www.biomedcentral.com/submit